Perspectives on testing for programming aptitude

Abstract

This paper discusses limitations of programming aptitude tests. The use of multiple-choice type questions, the testwiseness of the college graduate group, and the inclusion of questions of mathematical information tend to diminish the effectiveness of such tests as predictors of success in business programming. Timed tests favor the faster worker over the slower one who is often more accurate and of equal or better logical capability. The usual criterion of a test's validity is itself often not valid as a simulation of the true criterion but is accepted as the equivalent of the criterion because of its measurability and its appearance of plausibility. Comparative results of various occupational groups on the author's Aptitude Assessment Battery: Programming test are analyzed based on the scores of more than 11,000 persons tested, including more than 2,700 persons with work experience in programming in 317 companies, institutions, and government agencies.